Skill Orchestrator

A meta-skill that discovers, selects, and chains multiple skills to complete complex tasks that no single skill can handle alone.

Overview

Skill Orchestrator solves a key pain point: you have 50+ skills installed, but they can't work together effectively. This skill acts as a conductor - it analyzes your task, picks the right combination of skills, defines execution order, and manages data flow between them.

Features

• Automatic Skill Discovery - Scans installed skills and builds a categorized registry
• Smart Pipeline Selection - Matches your task to pre-built workflow templates
• Dynamic Orchestration - Constructs custom pipelines when no template matches
• Data Handoff Management - Handles transformation between skill outputs/inputs
• 8 Pre-built Templates - Covering WeChat, XiaoHongShu, Toutiao, financial analysis, and more

File Structure

skill-orchestrator/
├── SKILL.md                          # Core orchestration logic
├── scripts/
│   └── scan_skills.py                # Local skill scanner
└── references/
    ├── skill_registry.md            # All installed skills (auto-generated)
    └── pipeline_templates.md         # 8 workflow templates

Quick Start

Trigger Keywords

Activate this skill when you say:

• "全流程", "一站式", "编排", "流水线"
• "skill组合", "帮我用skill", "多个skill协作"
• "帮我用xxx和xxx"
• Cross-platform tasks like "research and publish", "search and create cards"

How It Works

1. Analyze - Decomposes your task into sub-tasks
2. Match - Finds the right skills from the registry
3. Orchestrate - Selects or builds a pipeline
4. Execute - Runs each step with proper data passing
5. Deliver - Summarizes results and output locations

Example Pipelines

WeChat Article → Publish:

agent-reach → wechat-mp-writer → humanizer → canvas-design → wechat-publisher

XiaoHongShu Note:

agent-reach → xhs-link-to-card-pipeline → xiaohongshu

Financial Analysis:

neodata-financial-search → xlsx → canvas-design

Refreshing the Skill Registry

When you install new skills, run:

python scripts/scan_skills.py --paths ~/.workbuddy/skills --output references/skill_registry.md

For WorkBuddy Users

This skill is designed for the WorkBuddy AI agent ecosystem. Place in your user-level skills directory:

~/.workbuddy/skills/skill-orchestrator/

License

MIT